BF Sports Analysis Rugby xP

BF Sports Analysis Expected Points Rugby xP

xP (Expected Points) is a concept that is frequently used in different sports but there is nothing about it in rugby union.

How it is used in Other Sports
In Basketball, the xP from a shot on the 3 point line is much higher than mid range. Hence the value of players such as Steph Curry have increased immensely. You can read a lot more about that in Kirk Goldsberry’s book, Sprawlball (and our review of the book here).

In Golf, they use a similar concept of ‘Strokes Gained’, where they take the average number of shots from a certain distance from the hole that professional golfers take. They then assess golfers by seeing if their next shot location is more than 1 less than their previous shot. With this, they can then assess what parts of each players games they are strong/weak at. E.g Driving, shots from 150-170 yards or putting. Read more about Strokes Gained in golf here.

PGA Shots Gained. Golf Rugby. BF Sports Analysis
Strokes Gained: Off the Tee. Credit: pgatour.com

Impact of xP in these sports
These statistics have had a huge impact in both sports, players take a lot more 3 points shots, shots near the basket, seek free throws and avoid mid-range 2s. Golden State Warriors who have one of the most prolific shooters in the league with Steph Curry won the NBA in 2015, 2017 and 2018 with this principled approach. 

In golf, players now have a much more accurate way of assessing their abilities to score from certain distances vs their peers. Through looking at what the top players were gaining shots on, they found that driving ability is much more consistent week-to-week than putting.

This meant that players who drive well are more likely to finish near the top of the rankings at the end of the year than a player who putts well. On the other hand, when a player putts well, that is when they are more likely to win a tournament. 

How does it work in Rugby?
The difficulty in rugby is that there are 30 players on the pitch and it is particularly difficult to quantify what each individual is contributing at any given moment to a score. There is also far less individual data in rugby, primarily due to the number of games each team plays being far less than basketball, soccer and rounds of golf. 

But it is more possible to look at it from a team point of view. We’ll use an example from Super Rugby Aotearoa(SRA), a match between the Crusaders and the Chiefs on the 28th June 2020. 

Note: This is looking at a reduced version of BFSA xP to explain the idea and also to preserve more accurate values for clients. For example, this is looked at in terms of the pitch as 4 zones (Own 22, Own 22-Half, Half to Opp 22, Opp 22) rather than the 30 pitch zones we actually use to increase accuracy. We are also assuming tries to be converted at 75% so a try is worth 6.5 points in this example. 

Chiefs have a scrum in their own half. The chances of scoring from a scrum in your own half is about 1 in 81 (1.2%). So 1.2% x 6.5 (value of scoring a try) = 0.08 xP.

The Chiefs pack won a penalty at the scrum. Well done to the Chiefs forwards. From Super Rugby Aotearoa, BF Sports Analysis can estimate the average line kick distances to the nearest 5m. We’ll leave out variable of the angle in this example. The estimated SRA average for penalty kicks to touch is ~33m.

The average distance a penalty kick to touch travels after 3 rounds of Super Rugby Aotearoa
Chiefs average kick to touch. Aaron Cruden. Expected Points. Rugby.
Aaron Cruden vs average kick to touch length

So the Chiefs pack have achieved, on average a lineout in the opposition half. Which gives them an 8.9% chance of scoring a try from (0.58 xP). The pack have gained 0.5 xP. They will have also reduced the chances of the Crusaders scoring if they get turned over since they are further up the pitch.

Aaron Cruden kicking to Touch. Will Jordan keeps it in. Value. xP. BF Sports Analysis. Rugby Analysis. Ireland. New Zealand.
Cruden goes for more

Aaron Cruden went for more in his kick to touch. If it went into touch in the opposition 22m, it would give the chiefs a 24.4% chance of scoring or 1.59xP. So the pack gained 0.5xP. Aaron Cruden if he hit touch in the opposition 22m would have added 1.01xP beyond the packs value. Similarily, if his kick to touch was less than 33m and the lineout was on the half way he would have decreased value.

You’re probably thinking, that is a pretty big impact for a kick to touch – one point difference. Forwards did all the hard work winning the penalty and will have to win the lineout before either the backs or forwards score and they only added 0.5xP. That is why our model for our clients uses more detailed pitch zones.

As it turns out, in this occasion, Cruden misses touch due to some great work by Crusaders’ full back Will Jordan keeping the ball in.

Will Jordan keeps the ball in

Crusaders now have an xP of <0.01 from a Kick Return in their own 22. You can debate either way if this is a Turnover Attack or Kick Return which effects the xP, you can even classify this a missed kick to touch but there is a very limited data set with that.

So well done Will Jordan, you’ve decreased the chances of the Chiefs scoring a try from 24.4% to giving yourself a <0.01xP attack from your own 22m.

So Will Jordan pretty much by himself has reduced the chances of Chiefs scoring average of 1.59 points. Now from this situation, with the Chiefs having a Kick Return in their own 22, their chances of scoring are still less than the chance of Crusaders turning them over and scoring from that turnover – I’ll save you the math on that one but it’s to let you know that the ‘game-state’ (Who has the better chance of scoring next) is still in favour of the Chiefs at this stage.

When Will Jordan kicks the ball out to touch. We end up where we were supposed to all along with a Chiefs lineout inside the Crusaders half with 0.58xP. So Will Jordan’s full action of preventing the ball go into touch and kicking it back out further up the pitch is on average worth 1.01xP.

Result of Will Jordan’s kick to touch

That’s cool. What does that allow us to do?
It’s going to be tough to use this in rugby when it is such an interdependent sport to rate individuals. Although in this case, we were able to narrow down this value to 2 individuals, that is often not the case in rugby. How do you attribute an xP value to a player hitting a ruck for example. In the NBA, they still struggle to attribute an xP to the guy passing vs the guy scoring accurately with only 10 people on the court to worry about.

Alongside the difficulty to rate players individually in rugby, I believe there is less purpose to rate individuals in rugby due to the transfers of players involving far less money than that of NBA or professional Association Football.

We can assess where as a team, you are not making converting in certain areas. You can see your actual points scored vs the expected points scored for Lineouts in general, lineouts in the opposition 22, scrums in certain areas, whether you are scoring off turnovers or kick returns. It can objectively label whether a team is good or bad at converting those situations into points versus average.

By that note, xP allows us to compare across situations much easier. Our average xP added per scrums is lower than our average xP added per lineouts, so lets work on scrums – but wait. There has only been 4 scrums/team/game in Super Rugby Aotearoa this year vs 14 lineouts, so really you should want to have a higher value added per lineout rather than scrum. This also why you wouldn’t use ‘% chance of scoring a try’ to compare different possessions.

That information can inform your training and your understanding of where you can improve.

See the example below for how the Blues have got on so far in Super Rugby Aotearoa goal kicking after 3 rounds. We can then use this xP value to compare our goal kicking to our lineouts, scrums and turnover attacks to see where we are strongest or weakest relative to our opposition.

We can look at the Blues goal kicking and how much above average they are for goal kicking vs locations

But…
There is obviously limitations to this data. Does it account for wet weather, wind, opposition etc…

BF Sports Analysis does try to mitigate and consider these elements in our analysis but not without flaw as the model develops.

As a wiser man than me once said, “all models are wrong but some are useful” and that’s where xP fits in rugby.


BF Sports Analysis collaborates with clubs, schools and organisations to make analysis more efficient, effective and enjoyable.

We do this by providing analysis services to save our clients time and upskill our clients to professional levels to ensure our clients are effective and enjoying how they use analysis. 

If your club wants access to the Super Rugby Aotearoa data and wants a level above the analysis provided in this article, please contact bfitzpatrick@bfsportsanalysis.com for more details. 

Alternatively, subscribe to our email list on the right hands side if you’re reading this on a desktop or below if you’re reading this on a mobile to get updates on articles such as this being released. Keep an eye out for a report on Super Rugby Aotearoa looking at how xP can be used to analyse a team.